The success of ARPANET, initially created in 1969 by the Department of Defense to link a number of leading research universities, spurred interest in further efforts at less-regulated systems of information exchange and facilitation of communication across organizational boundaries. But ARPANET was a private network which required its member universities to have a research contract with the DoD. The National Science Foundation wished to open the lines of communication further, and to allow other research universities to participate in the network.

In 1970, the NSF began CSNET, which was simply a single machine which researchers were able to use for email exchange between one another. Although it was successful and became a popular service, this solution was very limited. However, the intentions of the NSF went far beyond simply providing universally-accessable email for researchers. To further their agenda, they began work on what would later become NSFNET.

NSFNET was to be a high-speed network which would be open to all university research groups. Initial design and planning for the proposed network was begun in 1984. The network backbone was constructed first, connecting the six supercomputer centers owned by the NSF, located in:

  • San Diego, California
  • Boulder, Colorado
  • Champagne, Illinois
  • Ithaca, New York
  • Pittsburgh, Pennsylvania
  • Princeton, New Jersey

Each supercomputer was connected with a much smaller-scale LSI-11 microcomputer (called a fuzzball). These fuzzballs were then connected with 56kbps leased lines. Thus the backbone was formed on exactly the same network hardware configuration that ARPANET used. NSFNET, however, was unique in that it utilized the TCP/IP protocol, and thus it was the first TCP/IP WAN.

The NSF funded regional networks allowing a total of 20 research universities to communicate through NSFNET. The network became extremely popular and traffic exceeded the original design capacity in short order. The NSF charter specifically had prohibited commercial interests from funding or joining its network effort, but the NSFNET's traffic was quickly overrunning a 448kbps fiber optics channel leased from MCI and by 1990 had exceeded a 1.5Mbps T1. The NSF encouraged the formation of the nonprofit corporation Advanced Networks and Services, a joint effort of MERIT, IBM, and MCI, and eventually ANS took over NSFNET (which became ANSNET) and upgraded its 1.5Mbps bandwidth capacity to 45MBps.

Eventually, the NSFNET backbone was no longer needed do to the widespread proliferation of commercial service providers and their privately-built networks. In 1991 Congress passed a bill creating the National Research and Education Network, which was to run at gigabit speeds. The research functions and responsibilities of NSFNET were transferred to the new entity and the existing ANSNET network hardware was sold to America Online in 1995.

sources:
  • Tannenbaum, Andrew Computer Networks: Third Edition Prentice-Hall 1996.
  • Martin, Brown, DeHayes, Hoffer, Perkins Managing Information Technology (4th Edition) Prentice-Hall 2002.
(a node your library production)